Abstract
Clinical trials require participation of numerous patients, enormous research resources and substantial public funding. Time-consuming trials lead to delayed implementation of beneficial interventions and to reduced benefit to patients. This manuscript discusses two methods for the allocation of research resources and reviews a framework for prioritisation and design of clinical trials. The traditional error-driven approach of clinical trial design controls for type I and II errors. However, controlling for those statistical errors has limited relevance to policy makers. Therefore, this error-driven approach can be inefficient, waste research resources and lead to research with limited impact on daily practice. The novel value-driven approach assesses the currently available evidence and focuses on designing clinical trials that directly inform policy and treatment decisions. Estimating the net value of collecting further information, prior to undertaking a trial, informs a decision maker whether a clinical or health policy decision can be made with current information or if collection of extra evidence is justified. Additionally, estimating the net value of new information guides study design, data collection choices, and sample size estimation. The value-driven approach ensures the efficient use of research resources, reduces unnecessary burden to trial participants, and accelerates implementation of beneficial healthcare interventions.
Similar content being viewed by others
Explore related subjects
Discover the latest articles, news and stories from top researchers in related subjects.Avoid common mistakes on your manuscript.
Introduction
Unnecessary or poorly designed clinical trials waste research resources [1] and delay implementation of effective interventions. A well-conducted trial can also waste resources if the collected information is irrelevant to patients, physicians, or healthcare policy makers. Importantly, wasted research resources and implementation delays negatively affect patients’ well-being and the efficiency of healthcare systems [1]. Therefore, before clinical trials are performed, we should assess and prioritise them based on their potential value and impact [2, 3].
In healthcare, research priorities can be set using several methods, e.g., using burden of disease or a qualitative assessment of the potential research impact [4]. These methods can identify some important research areas, but they do not use formal methodology to assess whether research is justified and how to design the best clinical trial. Therefore, they are unlikely to make the most efficient use of available resources [4,5,6].
Furthermore, trials are usually designed to control the type I and type II errors when making conclusions about the primary outcome of the trial using a statistical hypothesis test [7]. These primary outcomes are usually selected using expert consensus [5], rather than assessing the relevance of that outcome to clinical and policy decision making. Furthermore, the trial sample size, a key element of trial efficiency [8], is computed to control error rates [9]. This can lead to large sample sizes that cause feasibility issues, require excessive time and money, put an unnecessary burden on patients, and delay implementation of effective interventions [10, 11]. This paper defines this approach as the error-driven approach to clinical trial design.
Critiques of the error-driven approach highlight a risk of publishing misleading research findings [12] and a propensity to interpret research findings incorrectly [12, 13]. These errors mean that research efforts and resources are wasted and clinical and policy decisions are misinformed [13]. Therefore, to improve research impact and the use of research resources, we must approach clinical trial prioritisation and design differently.
Thus, this paper describes an alternative approach to research prioritisation and trial design that evaluates the value that research can provide to clinical and policy decision makers. This approach can help research groups prioritise and design future clinical trials that make the best use of limited research resources and ensure that trial evidence supports decisions around the use and reimbursement of interventions.
Iterative research cycles: errors versus value
Healthcare research is an iterative process, where treatment effect estimates are contested or confirmed in successive studies. Throughout this process, researchers rarely claim certainty about a particular result and usually call for more research. Findings from trials add to the current evidence but uncertainty around the effectiveness or efficiency of interventions is rarely eliminated. However, clinicians and policy makers must make decisions, even in the face of uncertainty. Thus, we must determine if the remaining uncertainty justifies further research and how to design research that reduces uncertainty efficiently and effectively. Research should be an iterative cycle of designing studies, analysing the collected evidence in the context of what is known already, refining the research questions and designing future studies until we can make justifiable decisions about improving clinical practice.
Error-driven approach
Currently, the majority of clinical trials are designed using the error-driven approach [14]. All research starts with a research question (Fig. 1a). Using this research question, the design process proceeds with a systematic review of existing evidence from clinical trials, where feasible accompanied by a meta-analysis. Sometimes the results from the systematic review are combined with information on risks and benefits, patient values and costs to support clinical or health policy decision making [15]. These processes may be sufficient to guide decision making but lack of statistical significance for a treatment benefit in the synthesised evidence, coupled with evidence or an a-priori belief of a true benefit, often leads to a new clinical trial.
In the error-driven approach, trials are commonly designed by selecting a key outcome, known as the primary outcome, and fixing the trial sample size so that a statistically significant difference can be seen for this outcome when the trial results are analysed in isolation. Following this isolated analysis, the trial data may be added to the previous systematic review and meta-analysis and the research cycle starts anew.
In general, the primary outcome is selected using expert consensus [5], often considering feasibility, e.g., progression free survival is used in oncology trials to allow for shorter follow-up times. The sample size calculation is often based on currently available evidence about the baseline behavior of the primary outcome and expert specification of the “minimally important clinical difference” [16]. Using these two values, the sample size is set to control the type I error typically below a nominal level of 5% and the type II error below 10 or 20% [9]. Thus, the error rates, clinical judgement and an informal incorporation of existing information of treatment benefit are the key drivers of clinical trial design in the error-driven approach.
Functions of “net value” used in health decision sciences and health technology assessment: \(Net \, health \, benefit \left( {NHB} \right) = HB - \frac{1}{WTP}*C\) \(Net \, monetary \, benefit \left( {NMB} \right) = HB* WTP - C\) Where: (1)HB is the health benefit, ideally integrating life expectancy and quality of life (e.g. QALYs) (2)C is the total costs, including healthcare and non-healthcare costs (3)WTP: society’s willingness-to-pay in monetary units for one unit of health Example: A treatment has an estimated health benefit of 12 QALYs (HB = 12), a cost of $200,000 (C = 200,000) and society’s willingness-to-pay is set to $50,000/QALY (WTP = 50,000). Then: \(NHB = 12 - \frac{1}{50,000}*200,000 = 8 QALYs\) \(NMB = 12*50,000 - 200,000 = \$ 400,000\) |
Box 1.
Value-driven approach
The value-driven approach asserts that research requires substantial investment of time, resources and money and should only be undertaken when it generates value. The value of research can be calculated using two key concepts: an estimate of the value of healthcare interventions and a suite of methods known as Value of Information (VOI) methods [17]. (For an explanation of the abbreviations used see Table 1).
VOI methods are applicable irrespective of the method used to value the healthcare intervention. Nonetheless, we usually take a health policy making perspective and value interventions using a composite of health benefit and costs [17]. Health benefit is either measured using “hard” outcomes, such as the number of life years saved, or, more commonly, by combining quantity and quality of life into a measure known as quality-adjusted life years (QALYs) [18, 19]. Costs can include directly related healthcare costs alongside wider societal costs, such as productivity or leisure time loss [18]. Health benefits and costs are then combined into one of two composite outcomes: the net monetary benefit or net health benefit. Net health benefit (NHB) measures the number of health units saved by the interventions, while the net monetary benefit (NMB) is evaluated in monetary units, e.g. $ [20]. Both measures require an estimate of society’s willingness to pay (WTP) for one unit of health [21], which can be thought of as an “exchange rate” between health benefits and costs (Box 1).
Once the value of each intervention has been calculated, we can find the best intervention by considering which has the highest potential value. However, the available evidence on benefits and costs is uncertain, resulting in uncertainty about the intervention that maximizes value. Thus, VOI methods estimate the value of future research as the chance of making the wrong decision about the best intervention with the current level of evidence, multiplied by the benefit of changing the decision in settings where we would be wrong. Thus, the value-driven approach is based on understanding that a decision about the best intervention must be made following a trial and controlling the consequences and probability of incorrect decision making.
Figure 1b presents the value-driven approach, starting with a research question, data collection and evidence synthesis. The value-driven approach then estimates the value of each intervention using information and data on health outcomes and costs. This process uses methods from statistics, health economics and decision science to characterise the impact of uncertainty on the estimates of value. The current best intervention for use in clinical practice is the intervention that is expected to have the highest value [22]. VOI methods then formally assess whether the current evidence is sufficient to determine the best intervention [23].
To achieve this, we estimate whether the cost of undertaking additional research exceeds the value of the research (bottom Fig. 1b) [3]. We can also compute the value of alternative trial designs to prioritise the trial protocol with the greatest net value [24]. The value-driven approach then assumes that evidence collected in the trial will be analysed and interpreted alongside the current evidence to improve decision making following the trial. The value of each intervention can be estimated using the updated evidence and VOI methods can determine if further research is required. Thus, the value-driven approach is a full iterative research process.
Steps of the value-driven approach
The steps of the value-driven approach are summarised in Fig. 2 and Table 2. While these steps may seem cumbersome, recent methodological advances and software can facilitate the process [25,26,27]. We provide a clarifying example of the value-driven approach in Box 2.
Firstly, the value-driven approach determines the clinical or public health decision making problem that is relevant to the research question. Next, we summarise the available evidence using systematic reviews and meta-analyses [28], integrate evidence on benefits and costs using a decision model [15] and calculate the expected value of each intervention.
Following this, we use distributions around the input parameters to model uncertainty in the current evidence and propagate this uncertainty through the decision model using “probabilistic sensitivity analysis” (PSA) [15]. PSA determines the effect of input parameter uncertainty on the expected NMB or NHB. Using the PSA results, VOI methods can determine the chance and the consequences of making the wrong decision about the best treatment, i.e., the value of future research (Fig. 2 and Table 2). Strictly speaking, VOI is the “expected cost of the uncertainty” where ‘’cost’’ is expressed in foregone health benefit or monetary units. Foregone benefit, or potential lost value, refers to the benefit that could have been gained if a more “optimal” decision had been made.
A VOI analysis begins by calculating the value of eliminating all sources of parameter uncertainty, known as the expected value of perfect information (EVPI). EVPI is the upper limit on the value that can be generated from a future study collecting evidence about the model parameters. If the EVPI is low, then no future research should be proposed [29].
As a clinical trial is unlikely to estimate all parameters that are relevant to the decision, a VOI analysis proceeds by considering which outcomes should be included in the future study by identifying the parameters that would generate the most value if we were to gather more information about them. This is assessed by computing the value of eliminating uncertainty in a smaller group of parameters using the expected value of partial perfect information (EVPPI) [30]. EVPPI is computed for different groups of parameters and those with the highest value should be considered as study outcomes. This information about the study outcomes of interest, helps to select the most appropriate study design. For example, a VOI analysis can help us to answer the question: ‘’should we undertake a longitudinal cohort study to determine incidence or a randomised controlled trial (RCT) to determine treatment effect?”.
The final VOI design phase uses the expected value of sample information (EVSI) to determine whether a specific trial design would give value. To undertake this final analysis, EVSI must be scaled up by the number of patients who could benefit from the research results to compute the population EVSI (popEVSI). If the popEVSI exceeds the cost of the proposed trial, the trial has value. To facilitate this analysis, we compute the expected net benefit of sampling (ENBS), defined as the difference between the popEVSI and the study cost [3]. If ENBS is less than zero, then this trial is not valuable, and the current best intervention should be used in clinical practice. Finally, ENBS can be computed for different study protocols by changing the study type, inclusion/exclusion criteria and sample size, to determine the most valuable trial [24]. Given that there is a cost associated with enrolling participants, the value-driven approach enrols participants until the value of their data is smaller than the cost of enrolling them in the trial [24].
How the value-driven approach addresses challenges in trial design
This section outlines how the value-driven approach can offer solutions to challenges researchers have to deal with when developing and performing clinical trials.
Efficiency
Clinical trials are expensive and time-consuming, mainly related to the required number of trial participants. The value-driven approach assumes that trial data is analysed within the totality of evidence relevant to the policy decision. This can reduce the sample size, cost and patient burden.
The value-driven approach also improves trial design efficiency when multiple interventions are available. In the value-driven approach, decisions about which interventions to include are made by considering which interventions are key to determining the best intervention upon completion of the research process. The error-driven approach uses expert consensus to determine the trial interventions, which may exclude valuable interventions from the trial.
Finally, the time required for research and reimbursement decisions delays implementation, which can have health consequences as effective treatments are slow to reach patients (see Box 2). The value-driven approach addresses the foregone benefits of delayed implementation explicitly as a cost of further research. Furthermore, it ensures that information for decision/policy making is available at the start of the trial and can be updated following the trial. Finally, clinical trial design is optimised to support decision making. The downside of this approach is that in-depth analysis is required in the trial planning phase, which requires more time and resources.
Generalisability
Clinical trials are often criticised for their lack of generalisability [31] and their use of outcomes with limited relevance to clinicians, patients or policy makers [32]. The value-driven approach addresses these issues by ensuring that trial outcomes will support decision-making. The value of trials collecting evidence on alternative outcomes, i.e., short-term surrogate outcomes vs long-term outcomes, can be compared to their required resources. This would determine whether the additional information in the long-term outcomes is worth the increased complexity and cost. The value of a trial is also proportional to the number of people affected by the decision (Table 2). Thus, if the trial has limited generalisability, the value of the trial is limited. This supports the development of trials with wide inclusion criteria. Moreover, reducing the time between trials and their implementation ensures that trial information more closely reflects current practice.
Example of Value-driven approach in trial design Willan and Kowgier compared value of information methods to traditional power calculations [33]. Example: A randomized clinical trial (RCT) funded by the Canadian Institute of Health Research (CIHR) investigating early vs late external cephalic version (ECV) for pregnant women presenting with a fetus in breech position. Error-Driven Approach: Primary outcome: Non-Caesarean delivery Sample Size Calculation: The investigators of the trial used evidence from a pilot study (n = 116 in both arms, where the proportion of non-Caesarean deliveries in the early ECV arm was 35.3% compared to 28.4% in the late ECV arm) [34]. The minimally clinically important difference was determined to be an 8-percentage-point increased probability of a non-Caesarean delivery in the early ECV arm. The type-II error rate of the trial was set to 0.20 with a two-sided type-I error of 0.05. Thus, the trial had an 80% probability of correctly rejecting the null hypothesis if the treatments differed by eight percentage points or more and a 5% probability of incorrectly rejecting the null hypothesis if there was no difference between treatments. The sample size was calculated for a two-sample test for proportions, including a continuity correction to adjust for binary outcomes [35]. Sample Size: 730 patients per arm This large trial was successfully funded by CIHR and completed in 2008 [36] Value-Driven Approach: Estimating Value: The prior distribution of the incremental net benefit was estimated based on the pilot data (probability difference (41/116 – 33/116)) combined with the assumed societal willingness-to-pay of $1,000 to achieve a non-Caesarean delivery. To estimate the total number of patients affected by the decision, a time horizon of 20 years and annual North American incidence of 100,000 breech presentations was assumed Sample Size Calculation: A decision model was developed to estimate the effects of both strategies based on the published pilot data. Next the uncertainty around the decision was simulated and the expected value that can be gained by reducing uncertainty (EVPPI) was calculated. This expected value of new evidence minus the cost of collecting this information yielded the expected net benefit of further research (ENBS). The sample size that maximizes the ENBS was selected as the optimal sample size. Sample Size: 345 patients per arm The value-based approach would have resulted in a 52.7% reduction in trial sample size Efficiency: The required trial budget would be reduced from $2,836,000 to $1,604,000 (43.4% reduction) The expected net monetary benefit of the trial would increase from $179,000 to $736,383 (around 4 times higher) Note that the two approaches take different perspectives allowing the value-based approach to require a lower sample size than the 730 patients required to achieve statistical significance in the error-driven approach. The value-based approach does not aim to achieve statistical significance. Instead, it optimizes the trade-off between collecting more information, which is costly, and making an incorrect decision about the best treatment. The value-based approach considers that a decision can be made between two treatments, even if the difference between them on some clinically-relevant outcome is not statistically significant [22].This allows the value-driven approach to potentially increase trial efficiency to such an extent that justifies the added complexity of designing research with this approach [33]. Note that Willan and Kowgier [33] also consider a two-stage value-based approach which increases the expected net benefit of the trial still further to approximately 8 times higher than that of the error-driven approach. |
Box 2.
Validity
Clinical trials can have issues with validity as patients switch interventions, are lost to follow-up and do not follow protocol [1, 37]. While treatment switching and protocol adherence are issues for the value-driven approach too, we can assess the impact of losing patients to follow-up and we can value efforts to reduce loss-to-follow up, i.e., financial incentives for follow-up questionnaires [38]. Furthermore, trial simulations that consider these issues can define how they influence the value of the trial.
Feasibility
The value-driven approach directly considers the available budget, the time-taken to undertake the trial and the delay in widespread implementation [3]. Thus, the value-driven approach designs trials that are, by definition, feasible conditional on budgetary and time constraints. Conversely, the error-driven approach can lead to designs that are infeasible, i.e., requiring infeasible sample sizes in rare diseases [39].
Personalised and precision medicine
Precision medicine is becoming an important part of healthcare [40, 41] but causes methodological issues in trial design [42]. However, by focusing on supporting personalised decision-making, the value-driven method can offer alternative trial designs that are feasible and generalisable. Furthermore, novel value-driven methods are available to optimise the design of trials in precision medicine [43].
Emerging technologies
Finally, fast evolving technologies can mean that interventions are outdated before trial completion. The value-driven approach considers that trial evidence will be added to the current evidence, facilitating adaptive trials compared to the error-driven approach. We can assess the value of new interventions and compute the value of adapting the trial to include them. Thus, the value-driven approach includes flexibility that ensures evidence is relevant to decision-making, even in the face of emerging technologies and a changing research landscape.
Discussion
This paper proposes the value-driven approach as an alternative to the current error-driven approach for clinical research studies, focusing on clinical trial design although these methods are also applicable to observational studies. We now discuss project- and system-level barriers to the widespread implementation of the value-driven approach and highlight the potential benefits of the approach.
Time required for Trial Design
Under the value-driven approach, it takes more information and time to design a trial as the value of each intervention must be estimated using decision modelling. This requires, ideally patient-level, evidence on the interventions’ costs and benefits as well as their effect sizes. In contrast, the error-driven approach focuses on a key primary outcome for interventions that have been selected for the trial using expert consensus. Thus, the value-driven approach requires a wider literature search and different modelling and data synthesis methods. However, these analyses are required for policy making and thus, we can reduce the time for trial data analysis by including them at the design stage. The increased time and cost of the research prioritisation and trial design process will require additional funds. However, as the value-driven approach optimises the spending of research resources, the savings from efficient and effective trials are expected to recoup the cost of this design phase.
Data access
The decision modelling required in the value-driven approach and the final trial analysis should include all the currently available data. This includes data in aggregated form from the literature and patient-level data from previous trials. Accessing these data to design a new trial may be challenging and, if these data are not made available, then the value-driven approach could develop designs that result in inefficient use of resources. However, wider data access and the secondary use of trial data are increasingly used to improve the efficiency of healthcare research [44]. In addition, recent efforts by academic journals on data sharing (e.g. Plos ONE requirements for public data access[45]) can facilitate such data access, and further efforts by other academic journals would facilitate this. Thus, reusing data to improve trial design and to ensure that research effectively targets decision uncertainty should be a key part of this effort.
Expertise required for Trial Design
The value-driven approach requires collaboration with an interdisciplinary group of researchers, including trialists, statisticians, health economists, and decision modelers for the trial design. There is a lack of expertise among trialists and statisticians with VOI methods, in part because the methods have been challenging to implement [26]. However, recent research has focused on facilitating VOI analyses [26, 46, 47] reducing the barriers to implementation by increasing education and software [25, 48,49,50,51]. Including researchers familiar with cost-effectiveness and VOI methods in the trial design will increase costs, again, offset by the more seamless and efficient use of resources in the trial and its analysis. Specifically, this collaboration ensures that information required for cost effectiveness analyses can be collected in the trial outcomes.
Adaptive research questions
If funding is available for the design and conduct of a trial, challenges may arise if the VOI analysis indicates that the proposed trial is an inefficient use of research funding, e.g., an alternative study or a smaller sample size may be required. In this case, funding may need to be returned or repurposed. Flexible funding instruments would allow researchers to undertake valuable research, even when it was not originally proposed. To benefit fully from the value-driven approach, the current method of research funding where deliverables are pre-specified will require modification.
Status Quo in regulatory processes
Regulatory authorities worldwide have strict guidelines around the type of evidence that must be submitted to demonstrate treatment efficacy and safety. If a trial is developed as the basis of a submission to these regulatory bodies, innovative trial designs, such as those developed through the value-driven approach, may be limited by those guidelines. However, there is an increasing trend for regulatory authorities to become more flexible and acceptive of innovative and efficient trial designs (e.g., umbrella/basket trial designs) given the challenges facing the current regulatory landscape (e.g., personalized medicine, expedited access). Additionally, value-driven approaches can suggest expanding the data collection beyond the typical primary efficacy/safety outcomes (e.g., costs or quality of life), thereby strengthening the evidence submitted to a regulatory body.
Current research infrastructure
The current research infrastructure and publication culture support the error-driven approach, e.g., statistically significant results increase the chances of publication in high impact journals [52], and analysing trial results within a decision model is less well accepted. However, the error-driven approach has been heavily criticised [13] and journals are beginning to accept trials analysed using alternative methods [53].
In conclusion, the value-driven approach has advantages over the error-driven approach as research performed based on a value-driven trial design will collect data that are valuable to society and thus reduce research waste [54]. The value-driven approach can also justify a more streamlined implementation of interventions, which is particularly important when facing an urgent situation affecting a large number of patients. The value-driven approach can guide the choice of study type, inclusion/exclusion criteria, sample size, allocation ratio, and criteria for adaptive designs.
References
Chalmers I, Glasziou P. Avoidable waste in the production and reporting of research evidence. Lancet. 2009;274(4):86–9.
Claxton K, Posnett J. An economic approach to clinical trial design and research priority-setting. Health Econ. 1996;5(6):513–24.
Conti S, Claxton K. Dimensions of design space: a decision-theoretic approach to optimal research design. Med Decis Mak. 2009;29(6):643–60.
Fleurence RL, Torgerson DJ. Setting priorities for research. Health Policy. 2004;69(1):1-10.
Minelli C, Baio G. Value of information: a tool to improve research prioritization and reduce waste. PLoS Med. 2015;12(9):1–5.
Mooney G, Wiseman V. Burden of disease and priority setting. Health Econ. 2000;9(5):369–72.
Flight L, Julious SA. Practical guide to sample size calculations: Superiority trials. Pharm Stat. 2016;15(1):75–9.
Sakpal TV. Sample size estimation in clinical trial. Perspect Clin Res. 2010;1(2):67–9.
Chow SC, Liu JP. Design and analysis of clinical trials: concepts and methodologies Second edi. Hoboken: Wiley; 2008.
Zhong B. How to calculate sample size in randomized controlled trial? J Thorac Dis. 2009;1(1):51–4.
Bouter LM, Zielhuis GA, Zeegers MPA. Textbook of Epidemiology. 1st ed. Houten: Bohn Stafleu van Loghum; 2018.
Ioannidis JPA. Why most published research findings are false. PLoS Med. 2005;2(8):e124.
Amrhein V, Greenland S, McShane B. Scientists rise up against statistical significance. Nature. 2019;567(7748):305–7.
NIH U.S. National Library of Medicine. Available: https://clinicaltrials.gov. Visited May 2020.
Hunink MGM, et al. Decision making in health and medicine. integrating evidence and values. 2nd ed. Cambridge: Cambridge University Press; 2014.
Freiman JA, Chalmers TC, Smith H, Kuebler RR. The importance of beta, the type II error and sample size in the design and interpretation of the randomized control trial. N Engl J Med. 1978;299(13):690–4.
Claxton K, Ginnelly L, Sculpher M, Philips Z, Palmer S. A pilot study on the use of decision theory and value of information analysis as part of the NHS Health Technology Assessment programme. Health Technol Assess. 2004;8(31):1–103, iii.
Briggs A, Claxton K, Sculper M. Decision modelling for health economic evaluation. Oxford: Oxford University Press; 2006.
Loomes G, McKenzie L. The use of QALYs in health care decision making. Soc Sci Med. 1989;28(4):299–308.
Stinnett AA, Mullahy J. Net health benefits a new framework for the analysis of uncertainty in cost-effectiveness analysis. Med Decis Mak. 1998;18(2Supp):S68–80.
Neumann PJ, Weinstein MC. Legislating against use of cost-effectiveness information. N Engl J Med. 2010;363(16):1495–7.
Claxton K. The irrelevance of inference: a decision-making approach to the stochastic evaluation of health care technologies. J Health Econ. 1999;18(3):341–64.
Claxton K, Sculpher M. Using value of information analysis to prioritise health research: some lessons from recent UK experience. Pharmacoeconomics. 2006;24(11):1055–68.
Willan AR, Pinto EM. The value of information and optimal clinical trial design. Stat Med. 2005;24(12):1791–806.
Baio G, Berardi A, Heath A. Bayesian cost-effectiveness analysis with the R package BCEA. New York: Springer; 2017.
Kunst N, Wilson ECF, Glynn D, Alarid-Escudero F, Baio G, Brennan A, Fairley M, Goldhaber-Fiebert JD, Jackson C, Jalal H, Menzies NA, Strong M, Thom H, Heath A; Collaborative Network for Value of Information. Computing the Expected Value of Sample Information Efficiently: Practical Guidance and Recommendations for Four Model-Based Methods. Value Health. 2020;23(6):734–742.
Alarid-Escudero F, Knowlton G, Easterly C, Enns E. Decision Analytic Modeling Package (dampack). R package version 1.0.0. 2021. https://github.com/DARTH-git/dampack
Higgins JPT, Thomas J, Chandler J, Cumpston M, Li T, Page MJ, Welch VA, editors. Cochrane handbook for systematic reviews of interventions version 6.2 (updated February 2021). Cochrane, 2021. Available from www.training.cochrane.org/handbook.
Wilson ECF. A practical guide to value of information analysis. Pharmacoeconomics. 2015;33(2):105–21.
Felli JC, Hazen GB. Sensitivity analysis and the expected value of perfect information. Med Decis Mak. 1998;18(1):95–109.
Sharpe N. Clinical trials and the real world: Selection bias and generalisability of trial results. Cardiovasc Drugs Ther. 2002;16(1):75–7.
Chalmers I, et al. How to increase value and reduce waste when research priorities are set. Lancet. 2014;383(9912):156–65.
Willan A, Kowgier M. Determining optimal sample sizes for multi-stage randomized clinical trials using value of information methods. Clin Trials J Soc Clin Trials. 2008;5(4):289–300.
Hutton EK, et al. External cephalic version beginning at 34 weeks’ gestation versus 37 weeks’ gestation: A randomized multicenter trial. Am J Obstet Gynecol. 2003;189(1):245–54.
Fleiss J, Levin B, Cho Paik M. Statistical methods for rates and proportions third edit. Hoboken: Wiley; 2003.
Hutton E, et al. The early external cephalic version (ECV) 2 trial: an international multicentre randomised controlled trial of timing of ECV for breech pregnancies. BJOG An Int J Obstet Gynaecol. 2011;118(5):564–77.
Henshall C, Latimer NR, Sansom L, Ward RL. Treatment switching in cancer trials: issues and proposals. Int J Technol Assess Health Care. 2016;32(3):167–74.
Heath A, Manolopoulou I, Baio G. Estimating the expected value of sample information across different sample sizes using moment matching and nonlinear regression. Med Decis Mak. 2019;39(4):346–58.
Abrahamyan L, Willan AR, Beyene J, Mclimont M, Blanchette V, Feldman BM. Using value-of-information methods when the disease is rare and the treatment is expensive - the example of hemophilia A. J Gen Intern Med. 2014;29(SUPPL. 3):767–73.
Collins FS, Varmus H. A new initiative on precision medicine. N Engl J Med. 2015;372(9):793–5.
Maughan T. The promise and the hype of ‘personalised medicine.’ New Bioeth. 2017;23(1):13–20.
Senn S. Mastering variation: variance components and personalised medicine. Stat Med. 2016;35(7):966–77.
Fairley M, Cipriano LE, Goldhaber-Fiebert JD. PS 4-55 Optimal allocation of clinical trial sample size to subpopulations with correlated parameters. 41st Annual meeting of the society for medical decision making, Portland, Oregon, October 20–23, 2019. (2020). Med Decis Mak, 40(1), E1–E379.
Cheng HG, Phillips MR. Secondary analysis of existing data: opportunities and implementation. Shanghai Arch Psychiatry. 2014;26(6):371–5.
PLOS ONE. Data Availability. Webpage. Available: https://journals.plos.org/plosone/s/data-availability. Accessed 26 Nov 2020.
Heath A, Manolopoulou I, Baio G. A review of methods for analysis of the expected value of information. Med Decis Mak. 2017;37(7):747–58.
Heath A, et al. Calculating the expected value of sample information in practice: considerations from 3 case studies. Med Decis Mak. 2020;40(3):314–26.
Fenwick E, et al. Value of information analysis for research decisions—an introduction: report 1 of the ISPOR value of information analysis emerging good practices task force. Value Heal. 2020;23(2):139–50.
Rothery C, et al. Value of information analytical methods: report 2 of the ISPOR value of information analysis emerging good practices task force. Value Heal. 2020;23(3):277–86.
Strong M, Brennan A, Oakley J. How to calculate value of information in seconds using ‘Savi’, the sheffield accelerated value of information web app. Value Heal. 2015;18(7):A725–6.
Grimm SE, Strong M, Brennan A, Wailoo AJ. The HTA risk analysis chart: visualising the need for and potential value of managed entry agreements in health technology assessment. Pharmacoeconomics. 2017;35(12):1287–96.
Cristea IA, Ioannidis JPA. P values in display items are ubiquitous and almost invariably significant: a survey of top science journals. PLoS ONE. 2018;13(5):1–15.
Quintana M, Viele K, Lewis RJ. Bayesian analysis: using prior information to interpret the results of clinical trials. JAMA. 2017;318(16):1605.
Macleod MR, et al. Biomedical research: increasing value, reducing waste. Lancet. 2014;383(9912):101–4.
Funding
Dr Hunink receives Royalties from Cambridge University Press for a textbook on Medical Decision Making, reimbursement of expenses from the European Society of Radiology (ESR) for work on the ESR guidelines for imaging referrals, reimbursement of expenses from the European Institute for Biomedical Imaging Research (EIBIR) for membership of the Scientific Advisory Board, and research funding from the American Diabetes Association, the Netherlands Organization for Health Research and Development, and the German Innovation Fund. Anna Heath was funded through an Innovative Clinical Trials Multi-year Grant from the Canadian Institutes of Health Research (funding reference number MYG-151207; 2017—2020). Eline Krijkamp is supported by the Society for Medical Decision Making (SMDM) fellowship through a grant by the Gordon and Betty Moore Foundation (GBMF7853).
Author information
Authors and Affiliations
Corresponding author
Ethics declarations
Conflict of interest
No conflict of interest.
Additional information
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Rights and permissions
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.
About this article
Cite this article
Heath, A., Myriam Hunink, M.G., Krijkamp, E. et al. Prioritisation and design of clinical trials. Eur J Epidemiol 36, 1111–1121 (2021). https://doi.org/10.1007/s10654-021-00761-5
Received:
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s10654-021-00761-5